1 Introduction

The underrepresentation of ethnic minorities is a widespread phenomenon across many domains, including business leadership (Cook & Glass, 2014), political leadership (Karimi, 2021), academic research (Bradford & Perry, 2021), and even clinical trials (Sheikh, 2006). Unsurprisingly, the underrepresentation of ethnic minorities in Hollywood has long been at the center of debate, while Hollywood has been accused of whitewashing (Chow, 2016). Recently, for instance, a cast of mainly White actors portrayed Egyptian deities in Gods of Egypt (2016). Driven by social movements advocating for equality, the public is asking Hollywood to increase the representation of ethnic minorities (Low & Jackson, 2020). Following these calls, 2020 marked the first time that ethnic minority actors exceeded proportionate representation, driven by gains among Black male actors (Hunt & Ramón, 2021). Paradoxically, Hollywood has also been recently accused of blackwashing when, for instance, Black actors were cast in leading roles in the 2014 adaptation of the Broadway production Annie (1977).

While research has started to examine the effect of Black actors on US movies’ domestic box-office (Kuppuswamy & Younkin, 2020), we examine the effect of Black actors on US movies’ country-level international box-office. In an additional departure from the literature, heeding calls to conduct research on intersectionality (Gopaldas, 2013), we distinguish between Black male and female actors and separately examine their impact on the country-level international box-office of US movies released in 2012–2019. Drawing from social identity theory (Tajfel & Turner, 1979) and the intergroup contact model (Allport, 1954), we investigate how these effects are moderated by (a) actors’ star power, (b) the number of releases prior to release in the country, (c) the time-lag between worldwide release and release in the country, and (d) whether the country is emerging (vs. developed).

We consider this to be an important contribution for multiple reasons. First, the representation of ethnic minorities remains a topic on which, unfortunately, opinions are split. While movements such as Black Lives Matter stress the need for equality, regardless of race and gender, many countries are still politically polarized (Pew Research Center, 2014), suggesting that responses to Black actors may be mixed. Second, globalization has led to an environment where foreign markets constitute a large portion of sales (Griffith et al., 2017). This is especially the case in the movie industry (Eliashberg et al., 2006). Due to cultural differences, consumers in different countries may respond differently to Black actors. Third, by separately examining the impact of Black male and female actors, this contribution responds to calls for intersectional studies (Gopaldas, 2013). From a societal perspective, insights into what may improve audience responses to Black actors and the movies they star in are needed in that they could be acted upon to promote greater representation and, in turn, more equitable opportunities for actors of all ethnic origins, with vast benefits for society.

2 Theoretical background

We extend the literature on international box-office, which has neglected representation of ethnic minorities, and the literature on the impact of representation of ethnic minorities on domestic box-office, which has neglected international markets and intersectionality issues.

2.1 International box-office

There is a burgeoning literature on movies’ international box-office, which has focused on its product- and country-specific antecedents (Table 4, Appendix 1). Looking at product-specific antecedents, scholars examined star power (Griffith et al., 2014; Moon et al., 2016), previous releases (Elberse & Eliashberg, 2003; Griffith et al., 2014), and the timing of releases (Elberse & Eliashberg, 2003; Griffith et al., 2014, 2017; Moon et al., 2016; Wu et al., 2022). Looking at country-specific antecedents, studies investigated national cultural dimensions (Griffith et al., 2014), cultural differences and compatibility (Moon & Song, 2015; Moon et al., 2016), and economic development (Griffith et al., 2014; Moon et al., 2016). To the best of our knowledge, no study has examined the effect of the representation of ethnic minorities on movies’ country-level international box-office.

2.2 Ethnic minority actors and domestic box-office

The effect of the representation of ethnic minorities on movies’ box-office is only now receiving attention (Table 5, Appendix 1). Aumer et al. (2017) find that whitewashing may be beneficial, as, under some circumstances, audiences prefer White actors. Hermosilla et al. (2018) find that the Chinese preference for fairer skin is associated with the frequent casting of fair-skinned actors in movies targeting China. Kuppuswamy and Younkin (2020) find that US movies with multiple Black actors have better domestic box-office.Footnote 1 Focusing on directors, Karniouchina et al. (2022) find that movies directed by women and minorities fare no different domestically compared to other movies.

The reasons for these mixed findings are twofold. First, these studies neglect heterogeneity within ethnic minorities. Consumers may respond differently to members of ethnic minorities depending, for instance, on their gender (Gopaldas, 2013). Second, consumers’ responses to ethnic minority actors may be contingent upon movie and country characteristics. Research is thus needed to understand under which contingencies ethnic minority actors may enhance or dampen international box-office.

2.3 Intersectionality

Crenshaw (1991, p. 1244) coined the term “intersectionality” to describe “the various ways race and gender interact to shape the multiple dimensions of black women’s employment experiences.” In general, the intersectionality paradigm argues that societal treatment of members of minorities is not homogenous. Despite having attracted considerable attention across numerous fields (Gopaldas, 2013), in marketing, intersectionality has been applied primarily in studies of vulnerable (Saatcioglu & Corus, 2014), impoverished (Lee et al., 1999), and subsistence marketplace (Viswanathan et al., 2010) consumers, surprisingly neglecting the experiences of Black women, whose societal treatment often differs from that of Black men as their ethnic minority status intersects with their gender minority status. The literature on ethnic minority actors has not distinguished between male and female actors, which partially explains the inconclusive results. Hence, as societal treatment of Black women differs from that of Black men (Crenshaw, 1991), it is worth investigating whether and how audience responses to Black male and female actors diverge. We note here that this research further deviates from the narrow marketing literature on intersectionality, as it examines intersectionality from the supply side, i.e., service providers, in general, and actors in particular (vs. demand side, i.e., consumers).

3 Hypotheses

According to social identity theory (Tajfel & Turner, 1979), people classify themselves and others into social categories/groups. Group membership, such as membership in ethnic groups (Tajfel, 1978), guides intergroup behavior (Tajfel & Turner, 1979). To reach positive evaluations of one’s ingroup, people engage in social comparisons, exhibiting ingroup bias (i.e., elevating the ingroup while derogating the outgroup) and homophily (i.e., favoring intragroup relations over extragroup ones), often resulting in intergroup conflict. Overall, the concepts of ingroup bias, homophily, and intergroup conflict are key underpinnings of social identity theory (Jost et al., 2004). A well-documented intervention to improve intergroup relations is increased exposure to (contact with) the outgroup. According to the literature on intergroup contact (Allport, 1954), while initial exposure to the outgroup is stressful and imbued with suspicion (Bai et al., 2020; Ramos et al., 2019), positive responses can be gradually established with more exposure (Allport, 1954; Bai et al., 2020; Dovidio et al., 2003). Applying this reasoning to our context of investigation, we expect prior exposure to Black actors and, we add, Black people (the outgroup), in general, to result in better attitudes toward them and, in turn, better audience responses to movies casting Black actors.

3.1 Main effects

Black actors have been historically underrepresented in Hollywood (Kuppuswamy & Younkin, 2020). Extending developments in the literature on intergroup contact (Allport, 1954) that prior exposure improves responses to members of the outgroup, one could expect that historically underrepresented—and, therefore, less visible—Black actors may decrease US movies’ country-level international box-office. The gap is closing, however, and 2020 marked the first time that ethnic minority actors exceeded proportionate representation, driven by gains among Black male actors (Hunt & Ramón, 2021). We expect such changes in Hollywood hiring decisions, combined with the recent widespread media coverage of social movements opposing systemic racism, to have resulted in more (and better) recent exposure to Black actors. Against this backdrop, we expect Black male actors to increase movies’ country-level international box-office. Nonetheless, integrating developments in the literature on intersectionality (Crenshaw, 1991) with emerging evidence that Black women are still significantly underrepresented (Hunt & Ramón, 2021), we expect Black female actors to decrease movies’ country-level international box-office:

  • H1a(b): Black male (female) actors increase (decrease) a movie’s box-office in a country.

3.2 Moderation effects

3.2.1 Star power

Star power has been shown to drive box-office in previous research (Liu et al., 2014). Consumers are more familiar with stars (Griffith et al., 2017). Furthermore, as stars generate buzz, the media is more likely to cover them and their movies (Karniouchina, 2011). Hence, we expect that as a Black actor’s star power increases, superior prior exposure will result in better responses to the actor and, in turn, to the movie they star in:

  • H2a(b): The higher the star power of Black male (female) actors, the more positive (less negative) their effect on the movie’s box-office in a country.

3.2.2 Previous releases

According to the literature on the lead-lag effect (Kumar et al., 2011), more releases of a movie prior to release in the focal country should result in consumers’ greater exposure to the movie and, in turn, better box-office, as consumers in lag countries learn about the movie from consumers in lead countries (Dhar et al., 2012; Kumar et al., 2011). Hence, we argue that as the number of previous releases of a movie starring Black actors increases, greater prior exposure will result in better responses to the actors and, in turn, to the movie they star in:

  • H3a(b): The greater the number of previous releases in other countries, the more positive (less negative) the effect of Black male (female) actors on the movie’s box-office in a country.

3.2.3 Time-lag

Prior research has shown that the time-lag between a movie’s worldwide release and its release in the focal country reduces box-office (Elberse & Eliashberg, 2003; Griffith et al., 2014) due to the perishability of buzz and advertising (Elberse & Eliashberg, 2003; Griffith et al., 2017). As the time-lag increases, however, consumers may have more time to become exposed to the Black actors starring in the movie. A longer time-lag may, in fact, allow promotional messages to become more visible to consumers (Karniouchina, 2011). Hence, we argue that as the time-lag between worldwide release and release in the focal country of a movie starring Black actors increases, superior prior exposure will result in better responses to the actors and, in turn, to the movies they star in:

  • H4a(b): The longer the time-lag between worldwide release and release in a country, the more positive (less negative) the effect of Black male (female) actors on the movie’s box-office in the country.

3.2.4 Emerging country

Consumers in emerging (vs. developed) countries are less likely to be interested in other cultures (Skrbis et al., 2004) as well as to have the wealth and willingness to travel to experience other cultures (Cannon & Yaprak, 2002), both necessary conditions to increase one’s prior exposure to members of outgroups. Taken together, these insights hint at the fact that consumers in emerging countries where Blacks are not the majority ethnic group may have had less prior exposure to them compared to consumers in developed countries. Hence, we argue that, in emerging (vs. developed) countries, inferior prior exposure to Blacks will result in worse responses to Black actors and, in turn, to the movies they star in:

  • H5a(b): The effect of Black male (female) actors on a movie’s country-level box-office is less positive (more negative) in emerging (vs. developed) countries.

Figure 1 outlines the conceptual framework.

Fig. 1
figure 1

Conceptual framework

Full size image

4 Data

We collected data from the-numbers.com on Hollywood non-animation movies released in 2012–2019. For each leading actor (i.e., actor listed on a movie’s theatrical poster), we downloaded a close-medium shot from IMDb. We used machine learning to determine, for each actor, the ethnic group to which they belong (i.e., Asian, Black, Hispanic, or White). To do so, we chose kairos.com, a deep learning face and diversity recognition algorithm. We also used kairos.com to determine the gender of each actor. We provide the variables in Table 1 and descriptives in Table 6, Appendix 2.

Table 1 Variables
Full size table

Combining data (less observations with missing values) resulted in 15,119 movie-country observations (788 movies, 63 countries, Table 7, Appendix 3). We note that as we focus on international box-office, we excluded releases in the USA (Moon & Song, 2015). To allow for a clean test of the hypotheses, we also excluded majority-Black countries, i.e., countries where Blacks and/or Mixed Blacks constitute the majority ethnic group.

5 Estimation

Observing the box-office of a movie in a country requires the movie to have been released in that country. Hence, we run a Heckman sample selection model to predict a movie’s likelihood of release in a country as follows:Footnote 2Footnote 3

$$\mathrm{Likelihood}\;\mathrm{of}\;{\mathrm{Release}}_{ic}=\mu_0+\mu_1\mathrm{Black}\;\mathrm{Male}\;{\mathrm{Actors}}_i+\mu_2\mathrm{Black}\;\mathrm{Female}\;{\mathrm{Actors}}_i+\mu_3\mathrm{Emerging}\;{\mathrm{Country}}_c+\mu_4{\mathrm{Sequel}}_i+\mu_5{\mathrm{Remake}}_i+\mu_6{\mathrm{Real}-\mathrm{life}}_i+\mu_7\mathrm{Director}\;{\mathrm{Power}}_i+\mu_8\mathrm{Minority}\;{\mathrm{Director}}_i+\mu_9\mathrm{Critic}\;\mathrm{Review}\;{\mathrm{Score}}_i+\mu_{10}{\mathrm{Budget}}_i+\mu_{11}\mathrm{Star}\;{\mathrm{Power}}_i+\mu_{12}\mathrm{Major}\;{\mathrm{Producer}}_i+{\mu_{13}\mathrm{Cultural}\;{\mathrm{Distance}}_c+\mu_{14}{\mathrm{Indulgence}}_c+\mu}_{15}\mathrm{Female}\;{\mathrm{Actors}}_i+\mu_{16}\mathrm{Number}\;\mathrm{of}\;{\mathrm{Actors}}_i+\mu_{17}{\mathrm{Inst}}_{ic}+{\textstyle\sum_{\mu=18}^{22}}\mathrm{MPAA}\;{\mathrm{Rating}}_i+{\textstyle\sum_{\mu=23}^{31}}{\mathrm{Genre}}_i+{\textstyle\sum_{\mu=32}^{40}}\mathrm{Year}\;\mathrm{of}\;{\mathrm{Production}}_i+{\textstyle\sum_{\mu=41}^{102}}{\mathrm{Country}}_c+\alpha_{ic}$$
(1)

where \(\mu\)s are the parameters to be estimated, subscripts i are movies, subscripts c are countries, and αics are error terms. Inst is an instrument for the probability of a movie being released in a country. We use as instrument the average number of country releases for movies produced in the same year as the focal movie. The model is a probit model.

For hypothesis testing, we estimate the following model:

$${\mathrm{Box}-\mathrm{Office}}_{ic}=\beta_0+\beta_1\mathrm{Black}\;\mathrm{Male}\;{\mathrm{Actors}}_i+\beta_2\mathrm{Black}\;\mathrm{Female}\;{\mathrm{Actors}}_i+\beta_3\mathrm{Black}\;\mathrm{Male}\;{\mathrm{Actors}}_i\times\mathrm{Star}\;\mathrm{Power}\;\mathrm{of}\;\mathrm{Black}\;\mathrm{Male}\;{\mathrm{Actors}}_i+\beta_4\mathrm{Black}\;\mathrm{Male}\;{\mathrm{Actors}}_i\times\mathrm{Previous}\;{\mathrm{Releases}}_{ic}+\beta_5\mathrm{Black}\;\mathrm{Male}\;{\mathrm{Actors}}_i\times{\mathrm{Time}-\mathrm{Lag}}_{ic}+\beta_6\mathrm{Black}\;\mathrm{Male}\;{\mathrm{Actors}}_i\times\mathrm{Emerging}\;{\mathrm{Country}}_c+\beta_7\mathrm{Black}\;\mathrm{Female}\;{\mathrm{Actors}}_i\times\mathrm{Star}\;\mathrm{Power}\;\mathrm{of}\;\mathrm{Black}\;\mathrm{Female}\;{\mathrm{Actors}}_i+\beta_8\mathrm{Black}\;\mathrm{Female}\;{\mathrm{Actors}}_i\times\mathrm{Previous}\;{\mathrm{Releases}}_{ic}+\beta_9\mathrm{Black}\;\mathrm{Female}\;{\mathrm{Actors}}_i\times{\mathrm{Time}-\mathrm{Lag}}_{ic}+\beta_{10}\mathrm{Black}\;\mathrm{Female}\;{\mathrm{Actors}}_i\times\mathrm{Emerging}\;{\mathrm{Country}}_c+\beta_{11}\mathrm{Star}\;\mathrm{Power}\;\mathrm{of}\;\mathrm{Black}\;\mathrm{Male}\;{\mathrm{Actors}}_i+\beta_{12}\mathrm{Star}\;\mathrm{Power}\;\mathrm{of}\;\mathrm{Black}\;\mathrm{Female}\;{\mathrm{Actors}}_i+\beta_{13}\mathrm{Previous}\;{\mathrm{Releases}}_{ic}+\beta_{14}{\mathrm{Time}-\mathrm{Lag}}_{ic}+\beta_{15}\mathrm{Emerging}\;{\mathrm{Country}}_c+\beta_{16}{IMR}_{ic}+{\textstyle\sum_{\beta=17}^{21}}\mathrm{MPAA}\;{\mathrm{Rating}}_i+{\textstyle\sum_{\beta=22}^{30}}{\mathrm{Genre}}_i+{\textstyle\sum_{\beta=31}^{39}}\mathrm{Year}\;\mathrm{of}\;{\mathrm{Production}}_i+{\textstyle\sum_{\beta=40}^{101}}{\mathrm{Country}}_c+\varepsilon_{ic}$$
(2)

where βs are the parameters to be estimated, subscripts i are movies, subscripts c are countries, and εics are error terms. IMR is the inverse mills ratio from Eq. 1. Controls is the full set of controls in Table 1. Variables expressed in dollars (star power of Black female actors excluded) are winsorized. One potential concern is focal construct endogeneity. We offer a robustness check adopting a control function approach (Petrin & Train, 2010) (Table 8, Appendix 3). The results do not change.

6 Results

The results from the sample selection model are reported in Column 1, Table 2 (pseudo R2 = 47%). The instrument is significant (b =  − 4.84, p < 0.01).Footnote 4 Both Black male (b =  − 0.17, p < 0.01) and female (b =  − 0.18, p < 0.05) actors reduce a movie’s likelihood of being released in a country, effects that likely represent studios’ expectations of reduced box-office for movies with Black leads (Duke, 2014). In Column 2, we ran the box-office model including the key independent variables (R2 = 60%). In Column 3, we report the results from the model in Eq. 2 (R2 = 77%). We summarize the hypothesis testing results in Table 3.

Table 2 Results
Full size table
Table 3 Summary of hypothesis testing
Full size table

7 Discussion

This first study of the impact of Black actors on US movies’ country-level international box-office offers four contributions. First, by showing that casting Black actors affects international box-office, the study extends the literature on international box-office, which has overlooked the representation of ethnic minorities. Second, the study extends the nascent literature on the effects of the representation of ethnic minorities on domestic box-office (Kuppuswamy & Younkin, 2020), which has, conversely, overlooked international markets. In doing so, the study shows that mixed results in the literature can be explained by (1) heterogeneity within ethnic minorities and (2) movie- and country-level contingencies. Third, by separately examining the effects of Black male and female actors, this study answers calls to conduct research on intersectionality (Gopaldas, 2013). Last, the additional analysis we ran on majority-Black countries (Table 10, Appendix 3) answers calls to conduct studies on Black consumers, who have been traditionally neglected (Bradford & Perry, 2021).

This research is important from a practical perspective. The moderation effects indirectly support our reasoning that intergroup contact helps promote better audience responses to Black actors. While early analysts expected international markets to be inherently racist (Duke, 2014), we show that the actual problem may be a lack of prior exposure, a phenomenon that can be addressed by casting more Black actors in the first place. Our findings are useful for studios. Studios could consider delaying the release of movies starring Black male actors in the most relevant foreign markets, especially if these are emerging markets. They could also consider delaying the release of movies starring Black female actors in the most relevant foreign markets, anticipating launches in other countries. Interestingly, the marginal effects of the significant interactions of Black female actors (Fig. 2, Appendix 4) show that the negative effect of Black female actors disappears with numerous previous releases or long time-lags. Last, this research is important from a societal perspective. Hollywood movies are powerful sociocultural icons. Increasing the representation of ethnic minorities would improve the reception of movies with Black actors and promote more equitable opportunities inside and outside the movie industry, with vast benefits for society.

This study has limitations that represent directions for future research. First, we only examine Black actors. Although we run additional analyses using Asian and Hispanic actors (Table 10, Appendix 3), future work focusing on them would also be beneficial. Second, we do not look at intersectionality associated with age. In recent years, actors have become increasingly vocal about older women being disadvantaged in Hollywood. We would consider a future study of ageism to be a valuable extension. Third, we only look at whether a country is emerging (vs. developed). Although we run analyses using the Inglehart-Welzel classification of countries (Tables 8 and 9, Appendix 3), future work using other country characteristics would be useful. Last, by using an algorithm to classify actors, the choice of photos may have inadvertently introduced some bias. While we checked the face validity of the results using a subsample of actors, this should be borne in mind when interpreting our findings.